We were promised flying cars. You’d settle for someone answering the phone.
If you grew up watching The Jetsons, you probably remember Rosie. She handled the household chores, kept things moving, and had a few opinions of her own.
George had his push-button job. You have customers to help, quotes to prepare, emails to answer, and a phone that somehow always rings when your hands are full.
By the time you get home, there’s still paperwork waiting.
A flying car would be nice. But a little help getting through Tuesday? You could use that right now.
That’s where I think the conversation about AI for small business should begin.
You probably won’t want twenty different AI subscriptions
At the end of the day, I think many businesses will settle on one main AI assistant. Something like ChatGPT, Claude, or Gemini.
They’ll become comfortable with it, give it context about their business, and gradually connect it to more of the work they already do.
My expectation is that these companies will keep competing to handle more of your working day. One familiar place to explain what you need has obvious appeal.
That doesn’t mean one subscription already runs your entire business. Specialist software, extra services, and usage costs may still be part of the picture.
There are already practical building blocks for this: ChatGPT’s connected apps, Claude’s projects and knowledge bases, and Gemini’s connected apps for work accounts. Their capabilities and availability differ, but each offers ways to bring more context or connected information into the conversation.
The question is: what would you ask it to do first?
Give Rosie one job before handing over the house
You don’t need a complete AI plan before trying something useful.
Choose a task you understand. Something that repeats. Something where you can recognise a good result and catch a mistake.
Here are ten practical places to look. Some are drafts you can try today. Others, such as answering live calls or booking appointments, need connected services and a tested setup. A general chat subscription alone does not switch those services on.
1. Answer calls when you’re unavailable
An AI receptionist can be configured to answer routine questions, take caller details, and pass inquiries to your team.
Picture a customer calling after closing to ask whether you handle a particular type of renovation. The system can use your approved service information and collect details for a callback.
Your first step: Choose a small set of after-hours questions it should handle, then test those calls and the handoff to a person before routing customers to it.
2. Follow up on outstanding quotes
You’ve already done the visit and prepared the estimate. Following up keeps getting pushed to tomorrow.
AI can help prepare a friendly message using the quote details and your instructions. Tell it to preserve the price and scope, and leave out any discount you haven’t approved.
Your first step: Draft three follow-up messages, check them, and send them yourself.
3. Handle familiar customer questions
“Do you serve my neighbourhood?” “What are your hours?” “What should I bring?”
An assistant with approved business information can help answer those questions and pass unfamiliar ones to a person.
Your first step: Collect your ten most common questions, write down the correct answers, and decide who handles anything outside that list.
4. Make sense of a long email conversation
Ask AI to summarise the discussion, identify what needs your attention, and prepare a reply.
That gives you a starting point without reading the same twelve messages again.
Your first step: Try one email thread in a business-approved tool, then check the summary against the original.
5. Turn a project photo into a social post
You have the photo. You know the story. Explain it out loud.
What was the customer trying to accomplish? What did your team do? What would someone else find useful?
AI can help turn that explanation into a post for LinkedIn, Facebook, or Google Business Profile. Our AI social media service connects that work to planning, production, and review.
Your first step: Use one project photo you have permission to share and record a short explanation. Check that the draft describes what your team actually did.
For a fuller walkthrough, see What’s Cooking? How to Create Social Media Content with AI.
6. Prepare a proposal draft
Give AI your approved template and the details of the job. Ask it to organise the information into a first draft.
You still review the scope, pricing, and commitments before sending it.
Your first step: Draft the description of work for one service you regularly provide.
7. Reduce the back-and-forth over appointments
With a supported calendar connection and clear booking rules, an assistant can help customers find suitable appointment times. Our scheduling and reminders service is a place to explore that setup.
Your first step: Define one appointment type, including its duration, available hours, and time between bookings. Test a booking and a cancellation before opening it to customers.
8. Organise receipt information
Document-processing tools can extract dates, supplier names, and amounts from receipts for review, reducing manual entry.
Your first step: Test a small batch and check the extracted information before using it in your records.
9. Explain a business report
Ask AI to describe what changed in your sales or marketing report and suggest questions worth investigating.
It should distinguish what the numbers show from possible explanations. Our reporting and summaries service focuses on this area of business administration.
Your first step: Ask, “What changed, what needs attention, and what can’t we tell from this report?”
10. Turn your experience into a staff checklist
Explain a routine task the way you would explain it to a new employee. Ask AI to turn that explanation into clear steps.
Your first step: Have someone try the checklist and point out anything missing.
As your reference material grows, a knowledge assistant can help staff find and use approved business information.
“Once you see one useful task work in your own business, you start looking at everything else and asking, ‘Could it help with that too?’”
Set AI up where your work actually happens
This is something I explored in The Next Big Change in Your Working Day Might Be a Dot.
For my work, the experience changed when AI could work with the files and applications I already used on my computer.
There’s a difference between asking a general question and giving an assistant the materials needed to complete an assignment.
Your project photographs. Your service descriptions. Your brand information. The spreadsheet you need help understanding.
Browser-based tools can do substantial work too. The important part is the setup: which files and services the assistant can use, what it knows about your business, and what you allow it to do.
Installing a desktop app alone doesn’t complete that job.
Think of setting up a workspace for someone joining your team. You would show them where things are, explain the business, and tell them what needs checking.
Your AI assistant needs that direction as well.
Keep the business information current. When your services, prices, or processes change, the material it relies on needs to change with them.
For more on the gap between buying access and changing the work, read Houston, We Have a Productivity Problem: Why Your Basic AI Plan Is Grounding Your Star Employees.
An AI coach can help you get past the first hurdle
Think about working with a fitness trainer.
You can read about exercise. Watch videos. Buy the equipment.
Then someone stands beside you, shows you what to do, adjusts your technique, and helps you complete the first session.
Suddenly, getting started feels more manageable.
That’s how I see part of our role at Salient AI.
“Think of me and the Salient AI team as your AI tutor or coach. We can work right alongside you, then help regularly or check in when you need us.”
Through our AI consulting work, we can help choose a useful first task, prepare the business information, set up the appropriate tools, and work through the result with you.
You get to ask questions while doing something that matters to your business.
Why did it produce that answer? What information was missing? How do you ask for a change? What should you check before using the result?
Some people will want ongoing support. Others will get comfortable and call us when they want to try something new.
Both make sense. The goal is to help you use AI with more confidence and less friction.
Related reading: AI Is Failing in Your Organization. The Reason Isn’t What You Think.
The benefit can go beyond saving time
In our own business, we’re saving thousands of dollars using AI.
We’re also doing work that we probably wouldn’t have done before because the cost would have put it out of reach.
That second part matters.
When something becomes more affordable to produce, you can use it more often. You can test an idea, prepare another version, or give a smaller project attention that previously wouldn’t have made financial sense.
That’s our experience, rather than a promised result for every business.
On some tasks, the extra capacity can feel like having another pair of hands. The team still supplies the knowledge, makes decisions, and checks the work.
But once you experience that help on a real assignment, it changes how you look at the next one.
Start small enough that you’ll actually try it
You don’t need to hand over your phone system on day one.
Start with a post, an email draft, or an internal checklist. Something you can review before anyone else relies on it.
Try one task this week
- Pick the job. Choose one recurring task and describe what a good result looks like.
- Time your usual approach. Record how long a typical example takes.
- Give AI the right material. Use your template, facts and instructions in a tool approved for that information.
- Try it on three similar examples. Include your instructions, review and corrections in the time you record.
- Compare the whole process. Look at quality, effort and any added tool costs before deciding whether to repeat it.
This is a small experiment, not a promise of a particular saving. A fast draft that takes ages to fix hasn’t solved the problem.
At the end of the week, ask yourself:
- Was the result useful?
- Did the whole process take less effort?
- Could I repeat it?
- What would make the next attempt better?
One good experience gives you a reason to keep learning.
When that process works, workflow automation can help connect repeatable steps across your existing software.
A list of features is easy to ignore. Watching AI help with something you’ve been putting off is much harder to dismiss.
Let’s find your first Rosie moment
What’s the recurring task you wish someone would take off your plate?
Bring it to Salient AI. We can work through it with you, help set up the right tools, and show you how to keep using them.
You might want a coach beside you for a while. You might only need a few sessions to get moving.
Either way, we can start with one useful job.
The flying car can wait. Let’s get Tuesday under control.
References and further reading
Official product references
- OpenAI: Connected apps in ChatGPT — Connecting supported services for information retrieval and supported actions.
- Anthropic: What are projects? — Project workspaces, uploaded documents, and knowledge bases.
- Google: Use apps connected to Gemini with a work or school Google Account — Connected services and availability requirements for work accounts.
Related Salient AI articles
- The Next Big Change in Your Working Day Might Be a Dot
- What’s Cooking? How to Create Social Media Content with AI
- AI Is Failing in Your Organization. The Reason Isn’t What You Think.
- Houston, We Have a Productivity Problem: Why Your Basic AI Plan Is Grounding Your Star Employees
Written from Stephen Lau’s experience and point of view, with AI-assisted editing and illustrations. The illustrations depict imagined scenes.